AI: The future of iterative design
With the construction industry now more open to adopting new technology, after experiencing the benefits that digital ways of working can offer, what is next for the industry?
Here, Craig Johnson, Regional Account Manager – Steel Division at Trimble, explores the future potential that Artificial Intelligence could bring to the world of BIM.
BIM is an area of constant development, with many in the industry continually looking for ways in which we can further push the efficiency and productivity benefits that technology can offer to detailing, engineering, fabrication and construction workflows. Parametric design, or data-driven design as it is also known, is perhaps one of the most recent developments, with an increasing number of detailers and engineers adopting this way of working.
By using parametric design tools in conjunction with modelling software, designers are able to input the required rules, parameters and design algorithm and have the computer then generate the design output. A natural progression of this is the idea of computer-driven design. Here, you can push technology further by inputting the required parameters and allowing the computer to automatically generate various different design iterations, in an effort to determine the most optimum and efficient design solution.
With an increasing number of people now adopting parametric design within their BIM workflows, allowing the software and technology to have more power while still remaining in control of the inputs and outputs, the question is: what’s next?
While cloud-based software, such as Trimble Connect, is not necessarily new, it continues to be a great way of enabling a connected workflow, facilitating collaboration and communication between project teams. Essentially a huge, unlimited data storage facility, a project’s BIM model can be stored in the cloud, along with all of its associated drawings, schedules and documentation, which people can access, review and individually work on. However, what happens once a project has been completed? Often, the majority of this valuable data remains in the cloud, un-used and un-utilised by its owner. Yet, the rise of Machine Learning and Artificial Intelligence (AI) could change this.
Put simply, machine-learning is a form of AI, whereby it takes existing data and information and uses this to develop its own intelligence system; to learn and think in a way similar to humans and provide its own solutions. Typically, the more data a machine is exposed to, the better it will become at detecting and internalising patterns in said data and understanding and providing insights.
Within the BIM and construction industry, AI has the potential to successfully harness and utilise the significant amount of past project data currently unused, in turn helping to further improve and enhance our productivity and efficiency levels.
While every building structure is itself unique, detailing and modelling tasks can often be repetitive by way of nature and design. For example, different steel beams and columns and their various connection solutions are often commonly occurring within a design project. It is these similarities in data that offer the potential for automation; with a company able to utilise their experience and known good design choices from past projects to help automate, design and optimise the new.
Take the task of detailing a complex steel connection as an example. Through the use of AI and machine-learning, it is possible that BIM software may (in the future) be able to detect similarities and patterns between a user’s new model and their previously completed designs, automatically suggesting and recommending design solutions based on past projects. In this case, the optimum design could feature fewer welds, fewer bolts or even less steel, making it more cost-effective as well as easier to fabricate and assemble on site.
In addition to the time-savings that automated technology could deliver, both in terms of the initial detailing work and improved accuracy resulting in less rework, it could also contribute towards achieving the most optimum and efficient design. Imagine if AI technology was able to look at completed designs and categorise what worked well and what didn’t; taking this existing data and using it to improve the new. Collaborative platforms could even then feed fabricator and construction information, such as costs and time, into this, resulting in new BIM designs that are driven by fabrication and construction, in addition to design. What was easy to fabricate? What was easy to install? What was most cost-effective? What was most successful?
Ultimately, however, the success of AI in complex environments, such as BIM, depends greatly on acceptance. In order for the industry to benefit from such technological advancements, there has to be a sense of trust – trust and confidence in the solutions that such automated and machine-learned software suggest. Only then can we truly reap the rewards of our technological advancements.